Here are the Supplementary Data for the publication [1] (2024) https://www.biorxiv.org/content/10.1101/2024.05.28.596203v1.
Authors: Pauline Hermans, Matsvei Tsishyn, Martin Schwersensky, Marianne Rooman, Fabrizio Pucci
A per-protein summary for datasets D (derived from Mega [2]) and L (derived from S4038 [3]) are given in ./summary_D.csv and ./summary_L.csv respectively.
For each protein/protein domain, it includes information such as sequence length, depth of the multiple sequence alignment used, and Pearson and Spearman correlation with ΔΔG of the evolutionary scores and ΔΔG predictions.
For each mutation of the datasets D and L, experimental ΔΔG values, RSA of the mutated residue and evolutionary scores (LOR and WLOR) are in given in ./datasets/.
Sequences of proteins from dataset D and L are located in ./fasta/.
3D structures of proteins from dataset D and L are located in ./pdb/.
The multiple sequence alignments used in the publication and their relative sequences' weights are located in ./msa/.
File ./p_values.ods contains all pairwise p-values discussed in Supplementary Section 7 which compares performances measured in Sections 3.1, 3.3 and 3.5.
- All energy values are given in kcal/mol.
- We use the convention that destabilizing mutations have positive ΔΔG values.
[1] Hermans P., Tsishyn M., Schwersensky M, Rooman, M. & Pucci, F. (2024). Exploring evolution to enhance mutational stability prediction.
[2] Tsuboyama, Kotaro, et al. "Mega-scale experimental analysis of protein folding stability in biology and design." Nature 620.7973 (2023): 434-444.
[3] Zheng, Feifan, et al. "Assessing computational tools for predicting protein stability changes upon missense mutations using a new dataset." Protein Science 33.1 (2024): e4861.